Since upgrading to Proxmox 8, I’ve run into a regression with LXC backups when using an NFS share that resides on an NTFS volume.
SETUP
- Linux server with an external NTFS hard drive
- On that NTFS drive, I created an NFS share
- Proxmox mounts this NFS share fine and recognizes it as valid storage
WHAT WORKS
- VM backups succeed directly to this NFS share
- Before Proxmox 8, LXC backups also succeeded to this same share
- I can still restore successfully from `.tar.zst` backups if I copy them onto the NTFS-based NFS share after creation
WHAT FAILS (SINCE PROXMOX 8)
- Direct vzdump backups of LXC containers to this NTFS-backed NFS share fail with a permission error
- Example log (sanitized):
104: 2025-09-07 01:05:23 INFO: Starting Backup of VM 104 (lxc)
104: 2025-09-07 01:05:23 INFO: status = stopped
104: 2025-09-07 01:05:23 INFO: backup mode: stop
104: 2025-09-07 01:05:23 INFO: ionice priority: 7
104: 2025-09-07 01:05:23 INFO: CT Name: cardtrick
104: 2025-09-07 01:05:23 INFO: including mount point rootfs ('/') in backup
104: 2025-09-07 01:05:23 INFO: creating vzdump archive '/mnt/pve/ntfs-external/dump/vzdump-lxc-104-2025_09_07-01_05_21.tar.zst'
104: 2025-09-07 01:05:23 INFO: tar: /mnt/pve/ntfs-external/dump/vzdump-lxc-104-2025_09_07-01_05_21.tmp: Cannot open: Permission denied
104: 2025-09-07 01:05:23 INFO: tar: Error is not recoverable: exiting now
104: 2025-09-07 01:05:23 ERROR: Backup of VM 104 failed - command 'set -o pipefail && lxc-usernsexec -m u:0:100000:65536 -m g:0:100000:65536 -- tar cpf - --totals --one-file-system -p --sparse --numeric-owner --acls --xattrs '--xattrs-include=user.*' '--xattrs-include=security.capability' '--warning=no-file-ignored' '--warning=no-xattr-write' --one-file-system '--warning=no-file-ignored' '--directory=/mnt/pve/ntfs-external/dump/vzdump-lxc-104-2025_09_07-01_05_21.tmp' ./etc/vzdump/pct.conf ./etc/vzdump/pct.fw '--directory=/mnt/vzsnap0' --no-anchored '--exclude=lost+found' --anchored '--exclude=./tmp/?*' '--exclude=./var/tmp/?*' '--exclude=./var/run/?*.pid' ./ | zstd '--threads=1' >/mnt/pve/ntfs-external/dump/vzdump-lxc-104-2025_09_07-01_05_21.tar.dat' failed: exit code 2
WORKAROUND
- If I back up an LXC container to another NFS share on a Linux-native filesystem (ext4/xfs), the backup completes fine
- I can then manually move the finished `.tar.zst` archive over to the NTFS-based NFS share, and Proxmox restores from it without any issues
WHY I THINK THIS IS A REGRESSION
- VM backups still work fine to this share
- LXC backups *used to work* here prior to Proxmox 8
- The backup file itself is perfectly valid on NTFS, so there’s no reason Proxmox shouldn’t be able to create it there
- The failure appears tied to changes in how Proxmox 8 runs `tar` inside `lxc-usernsexec`
SUGGESTION
- Either restore the pre-8 behavior where LXC backups could complete to these shares, or
- Add logic so that if the target filesystem doesn’t support the required POSIX features, Proxmox stages temporary files locally but still writes the final archive to the NFS share
This would remove the inconsistency of “VM backups work, LXC backups used to work, but now fail.” Even if NTFS-backed NFS isn’t common in enterprise datacenters, it is real in mixed environments and labs where Proxmox is widely used.
Thanks for considering!
SETUP
- Linux server with an external NTFS hard drive
- On that NTFS drive, I created an NFS share
- Proxmox mounts this NFS share fine and recognizes it as valid storage
WHAT WORKS
- VM backups succeed directly to this NFS share
- Before Proxmox 8, LXC backups also succeeded to this same share
- I can still restore successfully from `.tar.zst` backups if I copy them onto the NTFS-based NFS share after creation
WHAT FAILS (SINCE PROXMOX 8)
- Direct vzdump backups of LXC containers to this NTFS-backed NFS share fail with a permission error
- Example log (sanitized):
vzdump backup status: backup failed
vzdump 104 --remove 0 --notification-mode auto --storage ntfs-external --compress zstd --mode stop --node t620 --notes-template '{{guestname}}'104: 2025-09-07 01:05:23 INFO: Starting Backup of VM 104 (lxc)
104: 2025-09-07 01:05:23 INFO: status = stopped
104: 2025-09-07 01:05:23 INFO: backup mode: stop
104: 2025-09-07 01:05:23 INFO: ionice priority: 7
104: 2025-09-07 01:05:23 INFO: CT Name: cardtrick
104: 2025-09-07 01:05:23 INFO: including mount point rootfs ('/') in backup
104: 2025-09-07 01:05:23 INFO: creating vzdump archive '/mnt/pve/ntfs-external/dump/vzdump-lxc-104-2025_09_07-01_05_21.tar.zst'
104: 2025-09-07 01:05:23 INFO: tar: /mnt/pve/ntfs-external/dump/vzdump-lxc-104-2025_09_07-01_05_21.tmp: Cannot open: Permission denied
104: 2025-09-07 01:05:23 INFO: tar: Error is not recoverable: exiting now
104: 2025-09-07 01:05:23 ERROR: Backup of VM 104 failed - command 'set -o pipefail && lxc-usernsexec -m u:0:100000:65536 -m g:0:100000:65536 -- tar cpf - --totals --one-file-system -p --sparse --numeric-owner --acls --xattrs '--xattrs-include=user.*' '--xattrs-include=security.capability' '--warning=no-file-ignored' '--warning=no-xattr-write' --one-file-system '--warning=no-file-ignored' '--directory=/mnt/pve/ntfs-external/dump/vzdump-lxc-104-2025_09_07-01_05_21.tmp' ./etc/vzdump/pct.conf ./etc/vzdump/pct.fw '--directory=/mnt/vzsnap0' --no-anchored '--exclude=lost+found' --anchored '--exclude=./tmp/?*' '--exclude=./var/tmp/?*' '--exclude=./var/run/?*.pid' ./ | zstd '--threads=1' >/mnt/pve/ntfs-external/dump/vzdump-lxc-104-2025_09_07-01_05_21.tar.dat' failed: exit code 2
WORKAROUND
- If I back up an LXC container to another NFS share on a Linux-native filesystem (ext4/xfs), the backup completes fine
- I can then manually move the finished `.tar.zst` archive over to the NTFS-based NFS share, and Proxmox restores from it without any issues
WHY I THINK THIS IS A REGRESSION
- VM backups still work fine to this share
- LXC backups *used to work* here prior to Proxmox 8
- The backup file itself is perfectly valid on NTFS, so there’s no reason Proxmox shouldn’t be able to create it there
- The failure appears tied to changes in how Proxmox 8 runs `tar` inside `lxc-usernsexec`
SUGGESTION
- Either restore the pre-8 behavior where LXC backups could complete to these shares, or
- Add logic so that if the target filesystem doesn’t support the required POSIX features, Proxmox stages temporary files locally but still writes the final archive to the NFS share
This would remove the inconsistency of “VM backups work, LXC backups used to work, but now fail.” Even if NTFS-backed NFS isn’t common in enterprise datacenters, it is real in mixed environments and labs where Proxmox is widely used.
Thanks for considering!
Last edited: